10

The question says it all. It's a commonplace that "airliner travel" for passengers is far and away the safest travel, on almost any metric (per mile, hour, per human-year, etc).

However, I've seen it said that for airline crew (who obviously fly ~once or even many times per day) the danger becomes as high as say "driving in Europe".

(I don't know how you quantify that - perhaps "deaths per year". Also, I would assume the implication is versus "consumer car drivers", not versus eg truck drivers.)

This question is strictly about airliner† travel, i.e. large 20+ seat aircraft flying scheduled routes for name-brand national and international airlines: I don't want to see the statistics polluted with GA, private planes, bush flying, Fedex, rock band aircraft crashes from the 70s, etc.

Any ideas on this? is it

  • Just a myth?

  • A distortion of statistics? (Something like "sure, crew obviously die more than passengers, but taxi drivers proportionately die spectacularly more than crew" - sort of thing?)

†["airliner" is defined end of story in the tags on this site, but I just thought I'd spell it out to avoid a rash of confusion]

DeltaLima
  • 83,202
  • 11
  • 272
  • 366
Fattie
  • 1,892
  • 1
  • 21
  • 41
  • PS I assumed this would have been fully addressed on here already; but couldn't find it. – Fattie Mar 27 '24 at 14:03
  • Origin of the comparison between driving and frequent flying of flight attendants – ROIMaison Mar 27 '24 at 15:15
  • 6
    The elevated radiation at high altitude is probably more of a health risk than a deadly crash. – Peter Kämpf Mar 27 '24 at 16:32
  • @PeterKämpf Indeed! I thought that might be the case (I didn't bother explicitly stating "crashes" since it seems implicit in the car case that radiation is not involved (except perhaps at https://en.wikipedia.org/wiki/Umling_La )) – Fattie Mar 27 '24 at 16:52
  • As @PeterKämpf says, the figures about crashes are likely to be a sideshow. The health impacts of increased radiation and other impacts of constant travel are likely to be much more significant. – Jack Aidley Mar 28 '24 at 10:41
  • Here's how I look at it. You're at more risk to be in a car crash than a plane crash, but a plane crash presents a higher risk of death or grave injury. So, a car crash is less likely to kill you but significantly more likely to occur. Does it even the balance? I don't know. – Deko Revinio Mar 28 '24 at 21:53

2 Answers2

9

Not an easy question. These statistics are scattered and hard to process.

One of the best summaries I've seen is here. It's not official, but the airline stats check against the original source. For GA, numbers have to be converted from 0.77 per 100k flight-hours to 35 per 1B miles at an average of 215 mph. For convenience, I'll use fatalities per trillion pax-miles or TPM. Fatality rates for crew tend to be a bit lower than for passengers, but not by far.

It comes down to:

  • 70 per TPM for airliners; numbers are pre-MAX and exclude 9/11
  • 110 per TPM for buses
  • 150 per TPM for trains
  • 7,300 per TPM for cars
  • ~35,000 per TPM for GA
  • 212,000 per TPM for motorcycles

That paints a picture of a 1:100 ratio of airline to car fatalities. That said, the average US car driver covers 37 miles per day. The average FA does 90 hours per month, which at 500 mph comes to 1,500 miles per day. So the claim isn't 100% true - a daily car commute in the US is still 2.5x more dangerous than a flight crew job. But the US averages more traffic fatalities than Europe with its shorter commutes.

This doesn't compare flying all day with driving all day, only flying as a job with a simple daily commute. That said, professional drivers do build up skill resulting in lower accident rates. Buses have great stats. Taxi stats have been polluted by modern taxis being primarily ride-hail services with many part-time drivers, but even they average a lower accident rate than car commuters.

The real myth is that road driving is an exception from the rule that proficiency takes time and effort to achieve. The statistics say it's not. Professional drivers are better and safer than amateur drivers by almost as wide a margin (1:70) as that between amateur GA pilots and professional ATPs (1:500). People who drive for a living are in fact better at it than the rest of us.

A confounding factor is that airliners, even ones from less-reputable builders or in less-reputable airlines, are designed, built and maintained to incomparably higher standards than GA airplanes. This definitely plays into that 1:500 ratio. Between cars and buses, though, the difference in build and service quality is less drastic.

GA's safety record is comparable to other extreme sports and on the low end of the extreme, similar to scuba diving. It's safer than skydiving, mountaineering, or motorcycling. But not in the range of common carrier safety. Part of the reason is lack of hardware redundancy, a larger part is a low skill threshold. It's a sport that can also be a means of transportation, not the other way around.

Therac
  • 26,835
  • 2
  • 73
  • 110
  • 1
    What happens more often: a fatal crash, or a flight being cancelled/delayed because the pilot died in a car crash during the commute to the airport? – DeltaLima Mar 27 '24 at 20:32
  • 2
    @DeltaLima Is the plane a [s]Calhoun[/s] Boeing? Is the car a Pinto? Is the airline based in a LDC? – Therac Mar 27 '24 at 20:40
  • Also a factor when comparing TPM - the typical bus might seat 50-ish passengers. There are smaller planes in scheduled service, but when you're looking at typical Boeing and Airbus airliners in scheduled service they are more like 100 - 400 passenger capacity. – manassehkatz-Moving 2 Codidact Mar 27 '24 at 21:15
  • 1
    @Fattie Among extreme sports, GA's mortality is far from the highest, and similar to scuba diving. But the right way to treat it, in my view, is as a sport, with its associated risks. – Therac Mar 27 '24 at 22:34
  • Why exclude 911? – Jpe61 Mar 28 '24 at 12:13
  • @jpe61 The 9/11 crashes (and any other bombing, hijacking, missiles, etc.) are arguably a "danger" of airliner travel. But my sense of the question - and I suspect others as well - is that this question is about safety issues due to weather, mechanical and human (e.g., pilot mistakes) factors independent of hostile actions from another party (whether terrorist or nation-state). – manassehkatz-Moving 2 Codidact Mar 28 '24 at 15:13
  • As far as "if GA is so dangerous, why not get rid of GA?", one of the key uses of GA is as a way for pilots to develop their skills. Arguably if GA was not an option then the only pilots would be ex-military, and we wouldn't have nearly enough pilots - from what I have read there is a chronic shortage even with GA. – manassehkatz-Moving 2 Codidact Mar 28 '24 at 15:15
  • @manassehkatz-Moving2Codidact acts of terrorism are an integral part of the risks involved with commercial aviation. They should not be excluded when considering how dangerous flying is. – Jpe61 Mar 28 '24 at 16:48
  • @Fattie why would 1:500 make you want to "end GA"? Are you implying it is your (or a regulator's) job to tell a GA pilot what risk (personally and knowingly) they are allowed to take? Maybe while we are at it, we should ban driving, because "too many" drivers end up killing themselves? There is a list of things we could do to make GA safer (getting pilots an IR rating, turbine engine(s!), anti-ice, real-time weather radar) but they all take money, and many GA pilots decide piston single VFR has an acceptable safety/cost balance. (Go be an aero engineer if you want to change the balance.) – Azendale Mar 28 '24 at 18:17
  • hi @Azendale I was really just making a hyperbolic or humorous point "FIVE HUNDRED X?!" However (even within that humorous context: ) note that as a card-carrying libertarian I couldn't care less if someone kills themselves rock climbing, flying etc. ... the point is if they kill people on the ground. Actually I was about to comment "Thank goodness when GA planes crash, it seems they very rarely hit suburbs, towns, houses ..." I'm not an aero engineer but I did work at the BMW-RR aero engine plant !! – Fattie Mar 28 '24 at 18:46
  • Scuba diving should be banned though - it's too dangerous – Fattie Mar 28 '24 at 18:48
  • I think this view fails to place enough emphasis on the elimination of variables (and variable behavior) that makes both airliners and buses "safer." Airliners are glorified buses, following well-established procedures and routes over and over, with no capricious variations that invite mistakes or mishaps. Individual car drivers take highly variable trips, changing lanes at will, amidst thousands of others doing the same thing. Likewise GA pilots are often maneuvering at will on arbitrary courses. – Oscar Mar 29 '24 at 00:59
  • @Oscar This would make for a decent separate question. That said, I would disagree: weather (for aircraft) and traffic (for cars) are much more important variations than the exact route being taken. – Therac Mar 31 '24 at 10:28
  • @Therac It's not about the exact route. It's about the SAME route and same procedures, with no off-the-cuff variation allowed. Check out top causes of GA accidents: https://generalaviationnews.com/2019/06/24/top-10-causes-of-fatal-ga-accidents/ I'd say at least half of those are essentially eliminated by strict adherence to (and repetition of) procedures. – Oscar Mar 31 '24 at 22:10
  • @Oscar I think this is worth a separate question. – Therac Mar 31 '24 at 22:24
3

From What percentage of airplanes are involved in a crash in their lifetime?, mean values derived from IATA 2014 statistics (IATA is the air carriers association):

  • IATA aircraft in the World: 23.000.
  • Flights per day per aircraft = 4.3.
  • A fatal accident after 3 millions flights.

You may read the linked question to see how we get to the following outcome:

  • A fictive airline with 320 aircraft, which is the size of an operator like Air France, faces one fatal accident each 6 years.

Now deriving new numbers from this result.

New assumptions:

  • A career is 40 years.
  • A crew works 5 days a week
  • 5 crews are required to cover 24 hours, thus each crew flies 4.8 hours a day. Thanks to @DeltaLima for providing this ratio.

From this:

$ \sf \small { \begin {array}{|l|l|l|} \hline \sf \small \text {Years between fatal accidents for the airline} & & \sf 6 \\ \sf \small \text {Daily probability for the airline} & \sf 1 / (6 \times 365) & \sf 4.57e-04 \\ \sf \small \text {Daily probability for a single aircraft} & \sf (4.57e-04) / 320 & \sf 1.43e-06 \\ \sf \small \text {Daily probability for a crew member} & \sf (1.43e-06) / (7/5) / (24/4.8) & \sf 2.04e-07 \\ \sf \small \text {Yearly probability for a crew member} & \sf (2.04e-07) \times 365 & \sf 7.44e-05 \\ \sf \small \text {Career probability for a crew member} & \sf (7.44e-05) \times 40 & \sf 2.98e-03 \\ \sf \small \text {Years worked before death} & \sf 1 / (2.98e-03) & \sf 1.34e+04 \end {array} } $

During their career, a crew member has a probability of dying in a crash of 0,298%. It means the probability is 100% after 13,400 years.

While I take a fictive airline of 320 aircraft, these final figures are mean values for any airline size. But the distribution of accidents is not flat, it depends also on local factors, e.g. how maintenance and training are delivered.

See also: What are the statistical probabilities of commercial aircraft accidents?

mins
  • 72,728
  • 27
  • 308
  • 441
  • Fascinating. Completely randomly googled factoid "The lifetime odds of dying in a crash for a person born in 2021 are 1 in 93" supposedly. (As always, statistics from any source are often totally nonsensical since very few people understand statistics but everyone thinks they do, but I just randomly googled that up, it's an interesting approach to stating the likelyhood.) – Fattie Mar 27 '24 at 16:56
  • Here's a probably reasonably reliable source for "deaths per 100,000 humans" (blue line) https://injuryfacts.nsc.org/motor-vehicle/historical-fatality-trends/deaths-and-rates/ Note though that the second chart, scroll down, deaders per mile, blue line, is drastically different. – Fattie Mar 27 '24 at 17:00
  • Hi @mins - indeed I ranted about that extensively in the question. – Fattie Mar 27 '24 at 17:00
  • I'm in my late 60s so this 5700 years looks like a good bet for me risk wise. – John K Mar 27 '24 at 17:56
  • @Fattie: "The lifetime odds of dying in a crash for a person born in 2021 are 1 in 93". I don't understand this figure. Overall people fly less than once a year, especially people born in 2021. Does it mean flying 50 times a year correspond a likelihood of more than 50/93 of dying? – mins Mar 27 '24 at 18:45
  • @mins - back up! that "The lifetime odds of dying in a crash for a person born in 2021 are 1 in 93" was in answer to your question about CARS. Crap, I forgot the link, https://www.injurylawyers.com/blog/the-odds-of-dying-in-a-car-crash/ (purely ftr, since it's not 1950, you can now just speak any setence like that in to the air and your device will give you the link!) Sorry for the confusion !! – Fattie Mar 27 '24 at 18:49
  • Re "So no crew from this operator could die before flying 6 years" --surely you didn't really mean to say that? – quiet flyer Mar 27 '24 at 18:56
  • 3
    @quietflyer: If for this airline fatal accidents have a mean period of 6 years, for the 6 first years, nobody died, the next day the whole crew and the passengers died. And don't call me Shirley :-) – mins Mar 27 '24 at 19:08
  • 1
    @mins I am trying to understand your logic in your calculation. You divide by a factor 8 to arrive at 5700 years. Are you assuming average 8 hour working days? I think that is extremely high. In staff planning we need approximately 5 FTE for 1 position in 24/7 operations. You need 3 people to work around the clock in 8 hour shift, but you also have to account for weekends, sick leave, holidays, training etc. – DeltaLima Mar 27 '24 at 20:26
  • @DeltaLima: Correct, 8hrs a day. I wasn't aware of the flight time, I've indicated that as well as other elements. It was just to estimate a maximum likelihood (or minimum period). So with 4.8 hrs, it's more than 9200 hrs between accidents, and 0.4% in a career. By refining other elements I'd not be surprised to get less than 0.1%. Thanks for the info. – mins Mar 27 '24 at 21:47
  • 1
    For pilots, FAA has a regulatory limit of 1,000 flight hours per year (FAR 121.481(f)). – user71659 Mar 28 '24 at 07:55
  • ".... It means the probability is 100% after 13,400 years." This statistical statement is obviously wrong and hints that you should rethink how this conclusion was derived. To make it more clear: Just because the chance of rolling a 6 on a die is 1/6, the chance to roll at least one 6 after 6 rolls is not 100%. After any finite number of rolls it will in fact never be 100%, instead it is 1-(5/6)^n, where n is the number of rolls. – GWD Mar 28 '24 at 12:24
  • @GWD: It's true for the dices, because the probability is not conditional, however if you have a prior knowledge there is a 6 each 6 rolls, then after 6 rolls the probability is 1. In my case I stated there is an accident each 6 years for a 320-aircraft fleet. – mins Mar 28 '24 at 12:50
  • @mins There’s no law requiring a crash every 6 years. If you say there’s an average 1 crash in 6 years, then it’s exactly the same as the dice: you roll a 6 once every 6 rolls on average. But not with 100% probability, not at all. – Chortos-2 Mar 29 '24 at 01:43